How therapeutic communities work: Specific factors related to positive outcome

Background: Therapeutic communities (TCs) are becoming increasingly widespread as a form of treatment for entrenched mental health problems, particularly addictions and personality disorders, and are equally used in educational, prison and learning disability settings. Despite growing evidence for their effectiveness, little research has been conducted to establish how TCs work to produce positive outcomes. We hypothesize that there are two specific factors that in combination contribute to TC effectiveness: the promotion of a sense of belongingness and the capacity for responsible agency. Although both factors are found in other therapeutic approaches and are important to the psychosocial aspects of psychiatric care more generally, we argue that their combination, extent and emphasis are unique to TCs. Material: Drawing on social and experimental psychology, we: (1) review research on a sense of belongingness and the capacity for responsible agency; (2) establish the mechanisms by which TCs appear to promote them; (3) draw lessons for TC practice; and (4) suggest why they may contribute to positive outcome. Discussion: A sense of belongingness is correlated with improved self-esteem and overall well-being. The capacity for responsible agency is central to behavioural change. TCs are typically used in fields where positive outcome requires both personal growth and behavioural change. We suggest that TCs are uniquely placed to demand such growth and change of their members because the sense of belongingness engendered by TC methods protects against the risks engendered by this demand. Conclusion: Empirically informed, evidence-driven research is necessary to understand how TCs work and how TC practice can be improved. This understanding may offer lessons for the improvement of psychosocial aspects of psychiatric care more generally.


Introduction
Psychotherapeutic approaches are generally understood to comprise both non-specific and specific factors that contribute to positive treatment outcome. Non-specific factors are present in all types of psychotherapeutic model. They include, for instance, the therapeutic alliance, the therapist's competence, and adherence to treatment protocols (Chatoor & Kurpnick, 2001). In contrast, specific factors are not present in all types of psychotherapeutic model. Specific factors refer to the set of distinctive techniques and interventions that characterize a particular type of psychotherapeutic model, distinguishing it from other types. They include, for instance, the use of cognitive restructuring techniques in cognitive-behavioural treatment, and transference interpretations in psychodynamic psychotherapy.
For some psychotherapeutic models, the factors specific to the model are well described and their contribution to positive outcome established. But such description and research is lacking for the therapeutic community (TC) model. This may be due to a number of factors. First, the countercultural aspects of TCs may lead to internal resistance to systematizing attempts (Manning, 2010). Second, there may be difficulty in standardizing across TCs given the heterogeneity of the approach (Hinshelwood, 2010). Third, TCs may be thought of as a container within which other psychotherapeutic models are implemented, rather than as a psychotherapeutic method in itself (Warren et al., 2003). Finally, TCs may be considered akin to a sociological intervention, comparable to a school or family, for which reductionist analysis is not appropriate or whose benefits are self-evident (Pearce & Autrique, 2010).
Nevertheless, there are agreed, general descriptions of many of the key principles, methods and features of TCs (see e.g. De Leon, 2000;Haigh, 1999;Rapoport, 1960). These descriptions are codified in the Community of Communities accreditation process, a UK quality assurance network that measures standards of good practice given these principles and methods (Paget & Keenan, 2006). These accreditation standards apply to both democratic and addictions TCs, the similarities between which far outweigh their differences (Haigh & Lees, 2008). Moreover, in the fields of health, prison and addiction, there is evidence that TCs produce positive outcomes (for reviews, see Lees, Manning, & Rawlings, 2004;Mitchell, Wilson, & MacKenzie, 2007;Prendergast, Podus, Chang, & Urada, 2002;Smith, Gates, & Foxcroft, 2006). In combination, such research suggests that, despite differences between particular TCs, and despite similarities between TCs and social institutions more generally, TCs nonetheless constitute a distinctive and unified type of psychotherapeutic model that can be profitably investigated via empirical methods and evidence-based research protocols.
This paper has two aims. First, and most substantially, we propose two specific psychotherapeutic factors in TCs that we suggest contribute to positive outcome: the promotion of a sense of belongingness and the promotion of responsible agency. Both factors can be found in other psychotherapeutic models to a degree; but we suggest that their combination, extent and emphasis are unique to TCs. We sketch basic research on the importance of belongingness and responsible agency for psychological health and well-being. We describe the distinctive set of techniques and interventions that TCs use to target them. Finally, we reflect on how these factors potentially operate powerfully in combination, together maximizing the possibility for change.
Second, in making this proposal, we aim to suggest an agenda for future empirical research. Given the relative uniqueness of these combined factors to TCs, it is natural to hypothesize that any particular effectiveness TCs can claim as against other psychotherapeutic models, relative to a particular psychological problem or population group, is due to their joint contribution. But, until research is undertaken to test this hypothesis, we cannot know that it is precisely these factors, in combination or singly, that contribute to the positive outcome of TCs. Quite generally, we thus hope that our proposal may spur empirically informed, evidence-driven research on TCs. More particularly, we hope our discussion spurs research on the extent to which a sense of belongingness and responsible agency, in combination or singly, is augmented through TC treatment in contrast to other psychotherapeutic models. In our view, this is a key step to achieving a better understanding of how TCs are effective and, correspondingly, how TC practice can be improved.
The promotion both of a sense of belongingness and of the capacity for responsible agency is likely to be important to the successful provision of the psychosocial aspects of psychiatric care quite generally, as they are important elements of healthy psychosocial functioning. Although our discussion focuses predominantly on TCs, we also therefore hope that drawing attention to these factors may hold a wider interest for all practitioners concerned with psychosocial care.

Belongingness
Belongingness constitutes a fundamental human motivation. The belongingness hypothesis claims that: human beings have a pervasive drive to form and maintain at least a minimum quantity of lasting, positive, and significant interpersonal relationships. Satisfying this drive involves two criteria: First, there is a need for frequent, affectively pleasant interactions with a few other people, and, second, these interactions must take place in the context of a temporally stable and enduring framework of affective concern for each other's welfare. (Baumeister & Leary, 1995, p. 497) Note that belongingness requires more than mere social contact: for the drive to belong to be satisfied, the contact must be frequent, stable over time, positive and expressive of mutual concern.
Importantly, reliable and valid measures of belongingness have been developed as clinical tools. Levett-Jones, Lathlean, Higgins and McMillan (2009) developed such a scale after a review of prior belongingness scales, and used it to measure belongingness in nursing students; their 34-item scale demonstrated high internal consistency and construct validity. Van Orden and colleagues (2008) constructed a questionnaire to measure perceived burdensomeness and belongingness in a study on the factors influencing suicidality, which demonstrated good internal consistency and construct validity. An example question on belongingness from this scale is: 'These days other people care about me.' They found an association between thwarted belongingness and suicidal ideation.

Basic research on belongingness
Belongingness has been found to be correlated with positive adjustment (hope) in adolescents (Davidson, Wingate, Rasmussen, & Slish, 2009). This correlation is particularly strong in peer-to-peer belongingness, which appears to have effects distinct from, for example, student-teacher belongingness (Ryzin, Gravely, & Roseth, 2008). Belongingness is also linked to rises in self-esteem; in answers from middleaged women, descriptions of events leading to raised selfesteem emphasized belongingness, while events leading to lowered self-esteem featured failures to connect (Baumeister, Dori, & Hastings, 1998). In general, belongingness tends to obey general motivational rules: thwarting it leads to intensification, satiation leads to decreased drive (DeWall, Baumeister, & Vohs, 2008). Lowered feelings of belongingness can lead to an increase in affiliative behaviours, including behaviour modification and acquiescence (Baumeister & Leary, 1995;Williams & Sommer, 1997) (although excluded individuals may seek affiliation with new social contacts rather than trying to re-establish old bonds (Maner, DeWall, Baumeister, & Schaller, 2007)).
Accepted people are generally better at self-regulation, but may not be willing to exert themselves in the interests of social acceptance (DeWall et al., 2008), possibly because of a satiation effect; why strive for something you already have? Conversely, social exclusion has a deleterious effect on self-regulation (Baumeister, DeWall, Ciarocco, & Twenge, 2005). In fact, socially excluded people tend to be less prosocial (Twenge, Baumeister, DeWall, Ciarocco, & Bartels, 2007) and more aggressive (Twenge, Baumeister, Tice, & Stucke, 2001;Warburton, Williams, & Cairns, 2006). These findings suggest a vicious cycle: social exclusion may lead people to engage in behaviours that make subsequent social acceptance less likely. Rejected people also have a tendency to derogate those by whom they feel they have been rejected (Bourgeois & Leary, 2001).
The experience of belonging has an impact on suicidality. Joiner (2007) hypothesized that the three factors necessary for suicide were the feeling of being a burden, thwarted belongingness, and a lack of hope. Various studies have attempted to examine this. In college students, Davidson et al. (2009) found that burdensomeness and thwarted belongingness predicted suicidality. In addition, they found a correlation between lack of hope and thwarted belongingness.
The absence of social bonds is linked to unhappiness, and social exclusion or rejection is associated with anxiety and depression. Research on guilt, loss and loneliness all point to the central part played by belongingness in wellbeing. Decreased belongingness is associated with increases in stress and mental health problems, as well as somatic illness such as heart disease (Hawkley, Burleson, Berntson, & Cacioppo, 2003); conversely, increases in belongingness lead to a decrease in health problems and an overall increase in happiness (for a review, see Baumeister & Leary, 1995). A psychotherapeutic approach that is able to promote belongingness is therefore likely to have a range of beneficial effects.

Belongingness and the TC environment and method
The essential elements necessary to promote a sense of belongingness are: (1) frequency of contact; (2) longitudinal stability; (3) positiveness of the contact; and (4) the presence of mutual concern (Baumeister & Leary, 1995). Many TCs are residential; even the least intensive 'mini-TCs' involve at least weekly contact over a series of several months, the most normal time span being 18 months (Pearce & Haigh, 2008). In other settings, TC treatment rarely takes less than a year, and can take considerably longer, particularly in prisons (Rutter & Tyrer, 2003). Members are normally expected to stay for the whole period of treatment, and member turnover tends to be slow. In residential TCs, members have contact with at least a subset of other members daily. Thus the requirement for (1) frequency of contact and (2) longitudinal stability appear to be met. This is illustrated in the core standards for TCs used by the Community of Communities in the UK (Paget, 2008), which require communities to meet regularly, and that 'community members spend formal and informal time together'; as core standards, these constitute essential features without which a programme is not considered to be a TC.
TCs require more than mere social contact. Mutual concern is promoted beyond the professional concern that might be shown by staff. TC relationships are characterized by challenge, support and shared responsibility. Many TCs employ a 'crisis' or 'special' meeting provision, where members can obtain support at any time the TC is operational (in residential TCs, any time day or night). Nonresidential TCs have developed a system of telephone and sometimes 'physical' support, where members call each other or visit one another to provide support when the TC is not meeting (Higgins, 1997). The need for contact to be positive overall is not required or advised in Community of Communities guidelines, but the nature of the support mechanisms in place indicates that mutual concern and supportive interaction are likely to be prominent in TCs that are operating well. What the core standards do require is that community members share responsibility for one another, make collective decisions that affect the functioning of the community, and consider and discuss their attitudes and feelings towards each other (Paget, 2008). Although mutual concern cannot be brought forth involuntarily, TC environments are explicitly structured to promote it. Regular meetings, formal and informal activities in which work, play and therapy are engaged in cooperatively, and a structure in which relationships between members are regularly considered and members are encouraged to share responsibility for each other both individually and as a community, all promote the development of mutual concern. These elements are central to the TC method. Indeed communalism, one of the four central elements of TC method according to Rapoport (1960, p. 62), which refers to a 'tight knit, interconnected, warm and intimate' network of relationships, represents an ideal environment to promote a sense of belongingness.
In principle, then, the requirements for (3) positiveness of contact and (4) the presence of mutual concern are clear aims of TCs. Of course, this aim is not always achieved in practice. Whether in fact the contact is positive for a particular person, in a particular TC, is no doubt variable, and is likely to depend on the state of the community and nature of the individual's personality and needs at the time. The fact that TCs vary in the effectiveness with which they operate over time is recognized (see e.g. Campling, 2001;Lees et al., 2004). Nonetheless, it is evident that (3) and (4) are explicit and clear aspirations of TCs, and routinely if not inevitably achieved in practice.
It seems clear that TCs promote belongingness, but it might be that this is true of a range of psychological approaches, rather than being particular to the TC method. In this context peer belongingness is of particular importance; mutual concern between peers is different from concern shown by a professional for a patient, student or client, and appears to promote hope in a way that care shown by a professional does not (Ryzin, Gravely, & Roseth, 2008). Psychotherapeutic approaches that promote peer belongingness will involve groups that are in contact frequently for a substantial period of time, and promote the development of mutual concern in a positive environment, in order to meet the above requirements (1)-(4). The only common psychotherapeutic interventions that fulfil these criteria are long-term groups prioritizing peer communication and concern over professional guidance. Longer-term groups that emphasize structured top-down communication at the expense of unstructured peer-to-peer interaction, such as the kind of psycho-educational methods found in cognitive behavioural therapy (CBT) and related approaches, will not meet these requirements. Long-term analytic, humanistic and self-help groups (such as 12-step programmes that meet frequently) may do so, but they are unlikely to promote concern and the feeling of responsibility for peers to the extent of TCs, and members rarely spend the amount of time in each others' company common in TCs. Thus, although the promotion of belongingness may be a factor in the success of some other approaches, it appears to be uniquely prominent in the TC method.
Having established that belongingness appears to be a therapeutic factor specific to TCs, we can ask what TC practice can learn from belongingness research, and whether the effects of belongingness tally with the effects of TC treatment.

What lessons can be applied to TC practice from belongingness research?
Suspension, which is widely used in TCs and involves the temporary withdrawal of privileges of membership such as attendance at meetings and use of support, is likely to intensify the need to belong and motivate the learning of new behaviours (DeWall et al., 2008). However, suspension is also likely to produce feelings of rejection (thwarted belongingness), which may promote suicidality (Van Orden et al., 2008), and poor self-regulation following perceived rejection may make it more difficult for the member to engage in prosocial behaviour and rejoin the community. When decisions to suspend a member are taken, the likelihood of heightening risks of suicidality and poor self-regulation should be carefully weighed against the gains in promoting the need to belong.
Similarly, many TCs discharge members if they are consistently unwilling to give up unhelpful coping strategies such as self-harm and drug and alcohol misuse, and this might also promote suicidality in the short term. This argues for TCs having a robust and accurate system for assessing willingness to change prior to joining the TC, possibly involving a trial of the TC environment such as through the use of a preparatory group (for a description of this approach see Pearce & Haigh, 2008).
There are limits to the power of possible suspension or discharge (Duvall, 2007). TCs should be realistic about the likelihood of change in response to the need to belong and be ready to call a halt to efforts to modify behaviour in the face of limited engagement from a member. If a member feels rejected they may then derogate -that is, denigratethe community or its members, which can make it more difficult to return (Bourgeois & Leary, 2001).
Scapegoating and associated failures of belongingness should be closely monitored. Research connecting such processes to poor outcomes including suicide (Van Orden et al., 2008) argues against excluding staff from informal or 'milieu' time when bullying or scapegoating may occur. Staff should be fully involved in informal time and scrupulous about bringing scapegoating and bullying processes to the attention of the community.
Staff in TCs should maintain a balance between interventionism and a laissez faire approach. Group processes that are not continually monitored can produce adverse effects, and if these lead to a decrease in the experience of belongingness, the effectiveness of the method is likely to be compromised. On the other hand, if staff become too active it will detract from the experience of peer support.
Belongingness research supports the use of votes for joining TCs, something that is widely used in democratic TCs. Only those whom the TC feel can belong are likely to benefit, and the act of voting a new member into the community starts the process of acceptance with a memorable ceremonial act.
TCs have been thought to rely on group (peer) pressure to achieve their effects (Bateman & Tyrer, 2002). If true, this might make the persistence of beneficial effects less likely once the member has left the community and the group pressure no longer operates, as well as predisposing to bullying and scapegoating those who do not conform. Whether or not beneficial effects from TC membership persist beyond the end of treatment, and whether significant harm is done to some members, are empirical questions. However, an alternative explanation from belongingness research is that, as opposed to group (peer) pressure, the need to belong, and the threat of not belonging, are motivating forces. In TCs the expectation that members desist from destructive coping mechanisms such as substance misuse or self-harm is a basic group norm. This, together with the use of contracts to focus attention on members' behaviour, and the support and psycho-education available to establish more healthy coping skills, are likely to prove powerful motivators to change. Lastly, group (peer) pressure differs from the process in TCs by virtue of the fact that membership is always fully voluntary; procedures and norms are specifically designed to be prosocial in regard to the wider community and all other members of the TC; potentially destructive group processes such as scapegoating and subgroup formation are controlled; and the process is designed to increase interdependence and eventual independence and responsible agency. All these aspects are monitored and when necessary modified by professionals.

Do the effects of belongingness tally with the effects of TC treatment?
Some of the effects of TCs are predicted directly by belongingness research and provide direct evidence of belongingness being a central aspect of successful TC treatment. An increase in sense of belongingness is associated with reductions in suicidality and aggression, and an increase in feelings of well-being. These are areas in which TCs appear to have an effect (Barr et al., 2010;Dietz, Connell, & Scarpitti, 2003;De Leon, Sacks, Staines, & McKendrick, 2000;Shuker, Sullivan, & Rivlin, 2010). Reductions in aggressive and disruptive incidents in prison TCs (Dietz, Connell, & Scarpitti, 2003;Newton 2010) tally specifically with research linking lack of belongingness to aggression and paucity of prosocial skills.
In addition to the specific effect of an increase in sense of belongingness, secondary effects are likely to be due to behaviour modification as a consequence of wanting to belong to the group. Reductions in deliberate self-harm and misuse of substances may fall in this category. These effects, along with the beneficial effects of increased belongingness on well-being while a member of the TC, predict lowered use of health resources; this is a reported outcome in TCs (Chiesa, Iacoponi, & Morris, 1996;Copas, O'Brien, Roberts, & Whiteley, 1984;Davies & Campling, 2003). Such research also suggests that gains persist beyond the end of treatment. This is likely to be due to the persistence of advantageous behavioural changes (promoted by increased self-efficacy, see below), along with modification of cognitive representations (such as 'I am not acceptable'). The answer to questions used to measure belongingness, such as 'These days other people care about me' (Van Orden et al., 2008) (noted above), is likely to reflect a relatively stable cognitive representation about the self as much as the current social reality in which a person finds him/herself after leaving a TC. Note too, of course, that in so far as TCs promote prosocial skills, they provide tools by which a person may in fact successfully change their social reality, reinforcing the modified cognitive representation.

Responsible agency
The promotion of responsible agency is central to any psychotherapeutic model that aims not simply at the acquisition of self-knowledge and reflective capacity, but at behavioural change. Behavioural change is crucial to improvement, let alone recovery, when the problem to be targeted involves actions and omissions: voluntary behaviour over which the person has at least a degree of control (Pearce & Pickard, 2010). Examples of such problems can be found within addictions, eating disorders and personality disorders, as well as anxiety and depression, where core symptoms or maintaining factors include actions and omissions. Recovery or improvement thus requires the person to take active steps to change unhelpful habits or entrenched patterns of behaviour: to choose to do things differently, and to find the will to execute this choice. Quite often, this requires a basic shift in attitude: behaviour that may feel compulsive and out of control comes to be recognized as in fact an expression of choice and subject to at least a degree of control. As a result, we can understand responsible agency as involving two basic capacities. The first is the cognitive capacity to reflect on one's behaviour, make decisions about how one wants to do things differently, form resolutions, and commit to change. The second is the practical capacity to see this resolution or commitment through: not to waver from the chosen course, or, if one wavers, to find a way to get back on track rather than sink into despair. Put crudely, the person 'takes responsibility' as we naturally put it for patterns of problematic behaviour.
Importantly, although there is no measure for 'responsible agency' itself, reliable and valid tools have been developed to measure a number of factors plausibly connected with it. Perhaps the most general measure of responsible agency is the Generalized Self-Efficacy Scale, which has a reliability (internal consistency) of .84 (Judge, Erez, Bono, & Thoresen, 2002). Generalized self-efficacy is defined as an estimate of the capability to mobilize motivation, cognitive resources and courses of action to exercise general control over events in one's life. Sample questions include: 'When I make plans, I am certain I can make them work' and 'I am strong enough to overcome life's struggles'. Judge et al. (2002) found a strong relationship between self-esteem, generalized self-efficacy, locus of control and neuroticism. Locus of control measures the tendency to attribute control and responsibility for events to one of three causes: oneself, powerful others and chance. Sample questions (Levenson, 1981) include: 'My life is determined by my own actions' and 'When I make plans, I am almost certain to make them work'. Finally, impulsivity scales may also provide helpful measures (Barratt, 1959). Impulsivity involves both a tendency not to plan for the future and a tendency to act on the spur of the moment (Patton, Stanford, & Barratt, 1995). Both tendencies clearly run counter to responsible agency, which involves forming resolutions and seeing them through. Sample questions include: 'I am future oriented' and 'I am a steady thinker' (reverse scored) as well as 'I do things without thinking' and 'I act on the spur of the moment'. Impulsivity is a core symptom of many of the disorders successfully treated by TCs: most obviously addictions and personality disorders, but also disorders involving, for example, binge-eating.

Basic research on responsible agency
Post-therapeutic well-being appears correlated with posttherapeutic narratives of therapy that emphasize service users' own agency as a force of change (Adler, Skalina, & McAdams, 2008). Such emphasis on the power of individual agency in all likelihood links to increased self-efficacy, which is known to predict the likelihood of continued smoking cessation as opposed to relapse, effective pain management, long-term control of eating and weight, and adherence to preventative health programmes (O'Leary, 1985). Increased self-efficacy has also been linked to lowered recidivism in sexual offenders (Pollock, 1996) and higher educational achievement (Zimmerman, Bandura, & Martinez-Pons, 1992). Self-efficacy is clearly essential to all rational decisions taken to initiate behavioural modification: if one believes one is powerless to do something, it is not rational to form an intention to try (Pearce & Pickard, 2010;Pickard, 2011). Such research further suggests that it is also essential to the capacity to sustain behavioural modification over time -not only to form resolutions to change, but to see these resolutions through (cf. Deci & Ryan, 1987). Finally, self-efficacy is also associated with higher self-esteem and improved affective state (cf. Deci & Ryan, 1987). Altogether, the evidence that it correlates with general well-being is high.
Internal locus of control predicts better ability to cope with stress (Krause & Stryker, 1984), which is also correlated with well-being. Note that stress is a strong predictor of relapse in addiction (West & Hardy, 2006), suggesting that the capacity to cope is important to sustaining behavioural change. High impulsivity ratings self-evidently correlate with failure to sustain behavioural change: they predict succumbing to immediate temptation, rather than acting now with future interests in mind.
Resisting the temptation to break resolutions and act against future interests requires the exercise of willpower. Willpower may be aided by self-efficacy and internal locus of control. Common sense suggests that it is easier to resist immediate temptation if one believes that one can. But there is also increasing empirical evidence for the existence of a discrete faculty of willpower that acts, to use a common metaphor, much as a muscle does. Research demonstrates that willpower is effortful to exercise, and its exercise depletes its strength in the short term, although it can build it up in the long term (for a review of the empirical literature, see Muraven & Baumeister, 2000). Importantly, the faculty of willpower is task general, not task specific (Muraven & Baumeister, 2000). Hence, exercising willpower in one domain affects one's short-term capacity to exercise it in a different, unrelated domain, until one has had the opportunity to rest the faculty, allowing it to return to normal levels. On the other hand, regularly exercising the power of the will leads to a long-term increase in the ability to do so, which will in turn lead to an improved ability to resist temptation to act against your long-term interests.
'Taking responsibility' demands the capacity to reflect on one's behaviour, make decisions about how one wants to do things differently, form resolutions and commit to change. It also requires that one has the willpower to stay the course -to see through one's commitment to change, despite temptations to revert to familiar habits and entrenched behavioural patterns. But, from a clinical perspective, it is natural to postulate that the motivation to employ these capacities and faculties to change one's life demands a prior affective commitment: one needs to believe in oneself and care about oneself enough, otherwise one will not be motivated to assess one's life objectively and embark on what is often a very difficult path towards change, hence the link between self-efficacy and self-esteem identified by Judge et al. (2002). Improved self-esteem may thus be a crucial causal agent to the promotion of responsible agency. That may be one reason why TCs are effective: in so far as a sense of belongingness improves self-esteem, it is an essential precondition of the effective promotion of behavioural change. We return to the interaction of these two factors below.

Responsible agency and the TC environment and method
A variety of psychological methods implicitly aim to encourage responsible agency and promote behavioural change. These include: • Motivational interviewing techniques, to engage service users and foster the desire to change (Rollnick & Miller, 1995). • Varieties of CBT, such as dialectical behaviour therapy (DBT) (Dimeff & Linehan, 2001), Systems Training for Emotional Predictability and Problem Solving (STEPPS) (Blum et al., 2008), and 'stop and think' training (McMurran, Fyffe, McCarthy, Duggan, & Latham, 2001), to help manage selfharm and other counterproductive behaviours.
These treatments presume that clients are capable of controlling their behaviour and deciding to change. But, unlike TCs, this presumption is tacit as opposed to overt. Consider, for instance, motivational interviewing. The clinician adopts a submissive, non-challenging stance, expressing empathy and encouraging the client to see the unwanted consequences of their behaviour as motivation to change.
Similarly, CBT and related therapies support behavioural change in a non-judgemental, enquiring way, explicitly designed to encourage self-reflection and problem-solving, but without risk of self-blame (for discussion of how responsibility can be encouraged but blame avoided, see Pickard, 2011). Emotional intelligence teaches clients to distinguish emotions and behaviour, to allow them to control their behaviour when in the grip of strong emotions, even if they cannot control how they feel. TCs are distinguished from these other methods because the promotion of responsible agency is explicit. The language of agency and responsibility permeates TC culture: members are not only encouraged but expected to see themselves in this light. Most clearly, this is evident when members challenge each other directly to 'take responsibility' for unhelpful behaviours and support each other to do so by means of contracts between a specific member and the group, with clear consequences if the contract is breached to hold the member accountable. Although TCs encourage members to write their own contracts in order to take ownership of the process of change, it is open to the community to impose a contract if it judges it necessary. This consists in an explicit expectation of responsible agency: a belief that it is within a member's power to change, and a demand that they do so, even if the member does not share this belief or want to take responsibility in this way. Consequences for breaching a contract typically involve community reflection and discussion of the reasons why the member did not adhere to it, and how they will do things differently next time, but can include suspension and, at the extreme, discharge.
Alongside the explicit use of contracts, TCs encourage responsible agency in a number of ways that collectively allow community members to take ownership of the group and thus responsibility for how it is run. In democratic TCs, new members are often accepted into the group through a voting procedure: the group has the power to decide whom to include. Although certain boundaries and rules are nonnegotiable in order to maintain good therapeutic standards and a strong, stable, safe community, many aspects of community life are devolved to the members, who may take collective decisions about what the boundaries, rules and ethos of their group are to be, but invariably have the responsibility to see that the boundaries and rules are respected. Although these decisions are about the group, this offers practice for the kind of responsible agency TCs aim to encourage in the individual: the group must reflect on what kind of group it wants to be, weigh evidence, hear different views and opinions, come to a decision and enact it. TC members are also responsible for numerous day-today aspects of community life, such as cooking, cleaning, hosting, administration and taking on responsible roles in the community. There is an expectation of responsibility for various roles and tasks within the community; if members do not take part in these responsibilities, they receive feedback from the community, along with support and encouragement to fulfill their responsibilities and roles.
Hence, with respect both to individual members and to the community, TCs encourage responsible agency via the general capacity to reflect, decide and enact change. Selfreflection, other-reflection and group-reflection are an important part of the community process, which promotes theory of mind skills and self-and other-empathy. Different members in the group offer different perspectives and desires, which need to be equally respected and weighed, and decisions taken in collective and fair ways. Finally, the community has the power to enact decisions taken. Throughout these processes, there is a standing expectation that members are capable of taking responsibility for themselves and the community. They are empowered to be reflective, responsible agents.
With respect to the capacity for willpower, various aspects of TC practice augment it. TCs often require members to stop unhelpful habits and problematic behavioural patterns, such as drug and alcohol use, binge-eating or deliberate self-harm; there is a standing belief that such control is possible, which may itself contribute to members' capacity to find the will to do so (Bandura, 1997). But alongside the requirement of abstinence is collective problem-solving and support to make this abstinence easier: replacement activities are identified, alternative, healthy coping strategies put in place, contracts are enacted and, despite negative consequences for breach, an attitude of support and compassion prevails. These aspects of TC practice target and augment willpower through the demand that it be exercised, the provision of better options, and consistent, stable, good regard, concern and support.

What lessons can be applied to TC practice from research on responsible agency?
The general capacity for reflection on one's mental state and behaviour involves theory of mind skills. Simply being a member of a reflective community that values such skills is no doubt beneficial, but TCs should consider empirical research on the nature of the development and improvement of mentalization and perspective-taking skills in order to improve the environment for learning. Note in this regard that MBT has consistently shown positive effects on borderline personality disorder (Bateman & Fonagy, 2008. No doubt, the kind of therapeutic environment offered by TCs should be expected to promote mentalization skills, but one way of ensuring this in personality disorder and related TCs would be to use Bateman and Fonagy's (2006) formal development of a MBT programme as part of the treatment.
TCs have high expectations of their members; arguably, that is part of why members actively work to take responsibility for their behaviour and meet those expectations. But failures and lapses are nonetheless frequent, and it is crucial that the imposition of negative consequences, which is part of how TCs hold members accountable, is both compassionate and fair. With respect to compassion, TCs must actively strive to retain an empathetic stance and clearly distinguish the practice of holding members responsible for behaviour from that of blaming them (cf. Pickard, 2011). With respect to fairness, the muscle model of willpower suggests that TCs must be careful to bear in mind the following: • The task generality of willpower. If a member is exercising willpower in one domain, that may temporarily affect their willpower in an unrelated domain, depleting them of resources that they and the community take for granted. • The possibility of genuine depletion. Willpower is a limited resource; although we should be cautious of too literal an interpretation of the 'muscle metaphor' it is clear that willpower can become run down. Especially during the period when abstinence from unhelpful behaviour or coping mechanisms is required, but positive alternatives have not yet been learned and cemented, the risk that willpower is severely taxed is high. TCs must acknowledge the reality of that risk and not impose expectations on members that they cannot be reasonably asked to meet.

Does what is known about the effects of the promotion of responsible agency tally with the effects of TC treatment?
There is good reason to hypothesize that the promotion of responsible agency is a key factor in the positive outcomes associated with TCs. By definition, behavioural change is central to improvement, let alone recovery, when the problem to be targeted involves actions and omissions, as with addictions and personality disorders. TCs appear to have positive effects on these disorders. They must therefore be promoting behavioural change. Given the degree and extent of the overt promotion of responsible agency within TCs, it is natural to conjecture that this is (part of) the cause of the positive effects that TCs have on these disorders. Given the evidence that increased self-efficacy, locus of control and willpower, alongside decreased impulsivity, predict the capacity to initiate behavioural change and sustain it over time (see above), one way of testing this hypothesis would be to measure change in these psychological attributes over the course of TC treatment as opposed to alternative modes of treatment. Such a study could provide evidence that TC practices that promote responsible agency do indeed effect improvement in attributes already identified as correlated with sustained behavioural change.

Combining belongingness with responsible agency: The unique power of TCs
There is evidence that TCs have positive outcomes with problems that are notoriously resistant to treatment, such as addictions and personality disorders. We want to conclude by suggesting that the power of the TC model depends on the unique combination of a sense of belongingness with the promotion of responsible agency.
We have already suggested that these factors may work in combination via self-esteem. A sense of belongingness improves self-esteem, which is then harnessed to initiate behavioural change: you need to care about yourself enough to genuinely want to take the difficult steps necessary to change things for the better. But the methods described above that are used by TCs to promote responsible agency and effect behavioural change carry risks -of members feeling a failure if they do not meet expectations set by themselves and the community, feeling rejected or punished when they are held accountable and face negative consequences, and hence of selfcriticism, self-blame, loss of hope, disengagement and relapse. Yet, when improvement or recovery requires behavioural change, it is difficult to see how it is possible without such methods. Belongingness, we suggest, protects against these risks. The employment of such methods within a community that is stable, meets frequently and is characterized by positive regard and mutual concern, not only motivates members to try to change, because they value the community and want the community to value them, but equally protects members against these risks by maintaining compassion and combating negativity. Support groups, analytic groups and humanistic groups promote belongingness, but less so responsible agency. The possibility of behavioural change is thus limited: support without challenge can potentially further cement bad habits and entrench problematic patterns. On the other hand, models that promote responsible agency outside of a community that creates a sense of belongingness risk causing self-criticism, self-blame, loss of hope, disengagement and relapse. That may be part of why methods such as CBT, whether individual or group, are so cautious about introducing the language of responsibility, and careful to preserve a non-judgemental attitude. By combining a sense of belongingness with the promotion of responsible agency, TCs offer a unique environment to support growth and change. Put crudely, they can demand responsibility because the demand comes from within a group of supportive peers: people who are all equal and people who all care.